Cloud & datacenter leader with 20+ years in infrastructure, from the terminal to the executive committee. I run large hybrid estates (Azure, AWS, GCP, VMware, Kubernetes) and turn them into numbers a CFO can read: €4.5M/yr footprint brought under FinOps governance at SAUR, ~€3.3M/yr saved on Renault's Azure platform. I own strategy, budgets and vendors, and I still ship the tooling myself, increasingly with LLM agents running on my own GPU infrastructure.
Hybrid multi-cloud estate for a €2.3bn / 12,000-employee water-utility group: Azure, on-prem VMware, AKS, Citrix, SQL Server, NetBackup, operated via an offshore managed-services partner.
- Run the group FinOps programme: first end-to-end view of the €4.5M/yr cloud + on-prem footprint, presented to the IT executive committee; €400–540k/yr savings pipeline, of which non-prod quick wins already verified at ~€79k/yr.
- Built an automated monthly FinOps reporting pipeline (Python, Terraform, Azure Pipelines) producing ~30 deliverables from billing exports: executive committee deck, chargeback for Finance, Databricks commitment burn-down, budget/forecast campaign, tagging and AKS audits.
- Flagged €15–18k/month of upcoming cost drift (expiring Fabric and VM reservations, idle environments) before it hit the invoice.
- Designed HULK, tag-driven Azure resource-lifecycle automation (140 resource groups) on a FinOps tag taxonomy enforced by Azure Policy; led a decommissioning campaign across 13 subscriptions / 5,500+ resources.
- De-facto Azure Landing Zone v2 platform owner: IaC standards (Bicep → Terraform with Checkov gates), 25-page developer runbook adopted group-wide, CI/CD agent pools as code (Packer images with Trivy + SBOM, zero-diff Terraform import of existing resources).
- Observability for legacy estates: built cvad_exporter (Citrix CVAD) and deployed/contributed to the open-source nbu_exporter (NetBackup), feeding Alloy + Mimir + Grafana.
- Governance of 12 AKS clusters and VMware (700+ orphaned VMDKs remediated); manage the offshore managed-services partner (scope, SLAs, escalations), reporting to the Infrastructure Director.
Digital arm of Groupe Renault: Cloud Foundry PaaS & Azure platform (~€11M/yr) powering connected-vehicle services for Renault & Nissan brands, under the "Build & Run the Business" service unit.
- Platform operations: Cloud Foundry (PaaS) + Azure infrastructure for connected vehicles (Renault & Nissan)
- Deployed centralised Grafana monitoring on AKS, reducing incident resolution time from 15 days to 48h max
- Containerisation and right-sizing programme → cloud costs −30% (~€3.3M/yr saved on an €11M Azure footprint)
- Recruited and managed a team of 7 cloud engineers plus contractors
- Rearchitected Big Data infrastructure; Azure resource provisioning and cost optimisation
Euronext-listed French music-tech group, 50+ countries; brands include TuneCore, Nuclear Blast, Naïve, Groove Attack. Hybrid estate: multi-datacenter on-prem + Azure datalakes + AWS Big Data.
- Managed hybrid infrastructure across multiple datacenters; server & network operations, vendor oversight and incident response
- Conducted infrastructure audits and proposed evolution plans to improve performance and resilience; organised on-call rotation
- Led initial AWS deployments for Big Data processing (existing data on Azure datalakes); recruited specialist profiles to support digital transformation
Micro-payment aggregation and operator billing group. Role expanded from Ops Manager to Group IT Director following the 2016 HIMEDIA acquisition (Allopass, MediaKioske/Audiotel) and creation of Mobiyo.
- Full infrastructure overhaul: designed and built multi-operator HA private cloud on OpenStack across 2 datacenters; later deployed Proxmox-VE platform for Mobiyo production
- Led offshore team then built and managed local team of admins & DevOps engineers; recruited senior technical profiles; ISO 27001 / PCI-DSS compliance
- Drove HIMEDIA acquisition integration (2016): merged Gibmedia and Mobiyo IT systems, defined new infrastructure, tooling, migration roadmap and agile project governance
- Budget ownership (IT + production); R&D on VOD service, VoIP and microservices debt-reduction initiatives; created AS203544 autonomous system
- In-house CDN + GeoDNS serving 8M+ readers, digital press distribution platform
- Full infrastructure overhaul: migrated to Proxmox-VE on dedicated OVH servers
- Database performance optimisation, application profiling, monitoring & alerting stack
- Release lifecycle procedures, security hardening of access
Fleet telematics leader, post-acquisition of WEBRASKA (French real-time cartography pioneer, powering Mappy). Responsible for production infrastructure continuity and evolution.
- Managed HA production infrastructure across main and remote sites for embedded fleet telematics services; rearchitected mail & SMS alert systems
- Deployed equipment monitoring stack, VPN Roadwarrior for remote access; supervised physical datacenter migration with zero service interruption
- Automated recurring production operations; implemented energy backup solutions; on-call management; on-site client interventions (Linux/network expertise)
- HA web & mail infrastructure, B2B hosting, VoIP (Asterisk administration), Prestashop as a Service
- Monitoring, virtualisation, LDAP/AD, DNS (BIND/PowerDNS), load balancing
- R&D on information systems: PHP CMS, multi-site applications, alumni management
- AIOps agent that heals, upgrades and writes post-mortems for a multi-node K3s cluster; 3-tier escalation (scripted skills → local 7B model → 27B model / Claude), human approval via Telegram. Target: ≥90% of incidents resolved without a human.
- Ollama / llama.cpp on RTX 5090 and 3090: benchmarked Qwen and Nemotron models on tool-calling; run Claude Code and Codex against a local 27B model (128k context) through a custom compatibility proxy.
- Built multi-agent systems on this stack (25-agent hosting operations platform, LLM-orchestrated event-driven services on K8s) and used Claude Code agents to deliver Hublow, an open alternative to Citrix CVAD (1,100+ commits in 3 weeks).
- Multi-node K3s with Cilium (BGP LoadBalancer), ArgoCD GitOps, Harbor, GitLab CI, Prometheus/Grafana, pfSense/HAProxy edge: runs this site and its live cluster metrics.
- Azure cost estimator for IaC (Go, in development): prices Terraform/OpenTofu plans at negotiated prices, with merge-request comments and drift detection. Contributor to nbu_exporter (github.com/cbijon).